In article <ib2vrb$3e...@lust.ihug.co.nz>,
Lawrence D'Oliveiro <l...@geek-central.gen.new_zealand> wrote:
> In message <8jd3m9fr5...@mid.individual.net>, Neil Cerutti wrote:
>
> > On 2010-11-03, Ben Finney <ben+pyt...@benfinney.id.au> wrote:
> >
> >> styles = [
> >> ("normal", "image", MainWindow.ColorsNormalList),
> >> ("highlighted", "highlight", MainWindow.ColorsHighlightedList),
> >> ("selected", "select", MainWindow.ColorsSelectedList)]
> >
> > Agreed, except cute stuff like putting those three items in
> > columns is just as bad.
> >
> > Code should be utilitarian rather than ornate, Shaker rather than
> > Victorian.
>
> Tufteâs concept of âchartjunkâ could perhaps be extended to
> âformatjunkâ or
> âprettyprintjunkâ.
Not at all. Tufte is all about making data easy to understand visually.
The chartjunk he rails about is colors, shadows, 3-d effects, etc, which
make a chart "look pretty" but don't add to comprehension. If you take
out the extra whitespace, you end up with this:
> >> styles = [
> >> ("normal", "image", MainWindow.ColorsNormalList),
> >> ("highlighted", "highlight", MainWindow.ColorsHighlightedList),
> >> ("selected", "select", MainWindow.ColorsSelectedList)]
which I think is more difficult to scan visually. In the first example,
I can see from the overall layout, without reading a single character of
punctuation, that this is a sequence of groups of three items. That's
harder to see in the second example.
Likewise, if I want to know what kinds of things are in the second field
of all the items, it's easy for me to visually scan down the second
column in the first example. To do that in the second example is much
more difficult.
If you want to talk about "codejunk", my biggest concern here is the
things in the third column. They all start with "MainWindow.Color", and
they all end with "List". Thus, those bits of text add bulk, and no
useful information. It would be even more readable by doing:
> >> styles = [
> >> ("normal", "image", Normal),
> >> ("highlighted", "highlight", Highlighted),
> >> ("selected", "select", Selected)]
which, of course would require some more code elsewhere to define the
new names. Whether this makes sense in the context of the overall
codebase is an open question, but taking the table in isolation, it
certainly improves the readability. Tufte would approve.
--
http://mail.python.org/mailman/listinfo/python-list
